Introduction

Oilseed rape (Brassica napus) is an amphidiploid species (2n = 38, genome AACC) derived from a more recent, medieval hybridization of the two closely related diploid species B. rapa (2n = 20, genome AA) and B. oleracea (2n = 18, genome CC) (U 1935). Oilseed rape is the second most important source of vegetable seed oil in the world, after soybean. It is the most important oilseed crop in China, Canada, Europe and Australia, where it delivers a high-quality, unsaturated nutritional oil that in Europe is also a major source of biodiesel. Since the 1970s the worldwide production area of oilseed rape has increased continuously to the current estimated level of about 30 million hectares in 2009/2010. The European Union produced rapeseed oil from a harvested area of 6.4 million hectares in 2009/2010, an increase of 32 % in comparison with 2005/2006 (Mielke 2010). This sharp increase was accompanied by a shortening of crop rotation cycles, sometimes down to 2 years in Northern Europe. One consequence of this has been an increase of Verticillium longisporum pathogen infection threatening oilseed rape production in Europe (Dunker et al. 2008).

In Europe Verticillium disease of oilseed rape became economically important in Sweden in the 1970s (Johansson et al. 2006a) and has increased strongly in Germany, France, Poland, England, Russia and the Ukraine since the 1980s (Karapapa et al. 1997). Although the yield-damaging potential of the disease is high, with up to 80 % reduction of single plant yield, the overall yield losses can vary considerably between years depending on climatic conditions (Dunker et al. 2008). Control of Verticillium diseases and other soil-borne and vascular diseases by fungicide treatment is not effective (Klosterman et al. 2009), hence the most promising long-term measure to control the disease is breeding of resistant cultivars.

The soil-borne phytopathogenic ascomycete V. longisporum can survive in soils for many years through the production of microsclerotia (Karapapa et al. 1997; Heale and Karapapa 1999; Steventon et al. 2002; Fahleson et al. 2003, 2004). After germination of microsclerotia in the soil, the fungal hyphae penetrate the epidermal cells of the roots and eventually enter the xylem elements. The spread and blocking of xylem vessels by fungal structures has been suggested to cause restricted water and nutrient transportation, leading to chlorosis and leaf yellowing, premature ripening and senescence as typical symptoms of Verticillium disease in oilseed rape (Eynck et al. 2007). Clear identification of symptoms on winter oilseed rape plants infected by V. longisporum is difficult until late in the growing season after flowering and during ripening (Supplementary Figure S1). In contrast, in greenhouse infection experiments applied in resistance screening programmes, when plants are infected by root-dip inoculation with high densities of fungal spores early in their development, rapid systemic spread of V. longisporum occurs in susceptible oilseed rape cultivars. This results in typical leaf yellowing symptoms and severe stunting of plants (Supplementary Figure S2) (Happstadius et al. 2003; Rygulla et al. 2007a, b, 2008; Eynck et al. 2009a).

Oilseed rape has a comparatively narrow gene pool which originated in a limited geographic region of southern Europe through spontaneous hybridizations between a restricted number of turnip rape (B. rapa) and cabbage (B. oleracea) genotypes (Allender and King 2010). Genetic diversity in elite breeding material has been further eroded by continuous selection for seed quality traits since the 1970s, in particular through introgression of double-low (00) seed quality (zero erucic acid, low glucosinolate content) essentially from two varieties as donors of superior quality (Allender and King 2010). As a consequence, 00-quality winter oilseed rape has a relatively low genetic diversity and lacks a broad spectrum of disease resistances. Extensive screening of diverse B. napus germplasm for resistance to V. longisporum under controlled greenhouse conditions revealed no resistance sources (Happstadius et al. 2003), and current European winter oilseed rape cultivars exhibit only low levels of tolerance against V. longisporum. On the other hand, V. longisporum resistance has been successfully transferred from the two progenitor species into resynthesized B. napus lines via the embryo rescue technique (Happstadius et al. 2003; Rygulla et al. 2007a, b). The quantitative resistance in these lines is predominantly derived from the B. oleracea C genome donor.

The genetic basis of resistance to the systemic spread of V. longisporum in B. napus and other Brassica species is unknown. A proteome expression study identified proteins up-regulated upon V. longisporum infection in roots, hypocotyls and leaves, including the typical pathogen defense-related enzymes β-1,3-glucanase, peroxidase, PR4 and endochitinase (Floerl et al. 2008). Histochemical analysis of B. napus and B. oleracea genotypes with different levels of resistance revealed that the resistance response to systemic V. longisporum spread is localized internally within the hypocotyls, and that the build-up of physical barriers, deposition of cell wall-bound phenolic compounds and lignin modification within the vascular system play a crucial role in the defense reaction (Eynck et al. 2009b). Similarly, various structural barriers that restrict invading hyphae of V. dahliae and other Verticillium species have been described in roots and stems of cotton, potato and tomato (Klosterman et al. 2009; Xu et al. 2011).

A number of studies characterising the interaction of Verticillium species have been performed in the model crucifer Arabidopsis thaliana, a close relative of Brassica crops. In contrast, there are only limited reports on the genetic basis of resistance to Verticillium ssp. in crop species (Klosterman et al. 2009). For example, the gene Ve1 conferring resistance to V. dahliae and V. albo-atrum race 1 has been identified in tomato (Solanum lycopersicum). Ve1 encodes a cell surface receptor that belongs to the extracellular leucine-rich repeat (LRR) class of receptor-like proteins. Arabidopsis plants transformed with tomato Ve1 showed resistance to V. dahliae and V. albo-atrum race 1, but not to V. longisporum (Fradin et al. 2011). In A. thaliana a single dominant locus on chromosome IV, VET1, was found to confer resistance to Verticillium infection (Veronese et al. 2003). Häffner et al. (2010) found that genes involved in flowering control on A. thaliana chromosome IV are localized within quantitative trait loci (QTL) for resistance to systemic spread, although further studies are required to determine if these regions coincide with VET1. Also nearby are nucleotide-binding LRR protein genes of the RPP5 super-family, with Toll/interleukin-1 receptor domains controlling resistance to the fungal pathogen Peronospora parasitica (Häffner et al. 2010).

In different A. thaliana ecotypes, high infection rates are not always correlated with severe disease symptoms such as stunting which can be observed in B. napus after greenhouse infections (Floerl et al. 2010; Veronese et al. 2003; Häffner et al. 2010; Steventon et al. 2001). Analyses of mutants deficient in hormone signaling suggest that the jasmonic acid and ethylene signaling pathways, but not salicylic acid signaling, are involved in resistance activation in A. thaliana (Fradin et al. 2011; Johansson et al. 2006b; Pantelides et al. 2010). In contrast, however, salicylic acid and its glucoside were detected in xylem sap from B. napus roots and hypocotyls, and increased in the shoots above the hypocotyls after infection by V. longisporum, whereas jasmonic acid and abscisic acid levels remained unchanged (Ratzinger et al. 2009). Since A. thaliana is not naturally infected by V. longisporum, however, the A. thaliana–V. longisporum pathosystem has limitations for simplistic transfer of information on putative resistance mechanisms to B. napus.

It is not known whether different races or pathotypes of V. longisporum exist. Nevertheless, genetic mapping in a doubled haploid (DH) population, with resistance derived from a resynthesized B. napus line originating from a cross between white cabbage (B. oleracea ssp. oleracea convar. capitata) and a winter turnip rape (B. rapa ssp. oleifera) as genome donors, revealed two major QTL for V. longisporum resistance on chromosomes C4 and C5. After infection with either a mixture of five Swedish V. longisporum isolates or a single-spore isolate, respectively, these two QTL consistently exhibited significant effects on resistance in multiple greenhouse environments (Rygulla et al. 2008).

The objective of the present study was to identify genomic regions involved in V. longispourm disease development and resistance in different B. napus genetic backgrounds. To achieve this, a QTL mapping approach was applied to identify QTL that are stable across environments and breeding populations, and to develop markers to pyramid complementary resistance loci from resynthesized B. napus and commercial oilseed rape lines. Furthermore, QTL analyses of phenolic compounds in hypocotyl tissues were performed to determine their relevance in the V. longisporum disease reaction and resistance in B. napus.

Materials and methods

Resistance donors and mapping populations

A total of 214 homozygous DH lines of the population ExR53-DH were used for genetic mapping. This DH population was produced by Saaten Union BioTec GmbH (Leopoldshöhe, Germany) (Radoev et al. 2008; Basunanda et al. 2010) from a cross between the inbred line Express 617, derived from the partially V. longisporum-resistant German winter oilseed rape cv. Express, and R53, a resistant, resynthesized rapeseed line. R53 originates from an interspecific hybrid between B. oleracea ssp. oleracea convar. acephela (kale) and B. rapa ssp. pekinensis (Chinese cabbage). Four other DH populations used for marker validation, SW08-190001, SW08-190002, DSV-1575 and DSV-1605, were produced by the plant breeding companies Lantmännen SW Seed AB (Svalöv, Sweden) and Deutsche Saatveredelung AG (Lippstadt, Germany) using genetically divergent resynthesized B. napus accessions as resistance donors. Details of the pedigree and resistance donors of all populations are listed in Supplementary Table S1.

Resistance screening

Seedlings of the DH lines and parents were inoculated with a V. longisporum isolate mixture in four separate greenhouse experiments during 2009 and 2010 at Göttingen University. A mixture of the German V. longisporum isolates VL40 and VL43 (Zeise and von Tiedemann 2002) in equal densities of 1 × 106 spores/ml plus mycelium fragments each was used for the inoculations. Each experiment included a total of 20–24 inoculated and 10–24 non-inoculated plants for each of the DH lines. Furthermore, 20–24 inoculated and 10–24 non-inoculated plants each of the susceptible B. napus cultivars Falcon and Laser and the partially resistant cultivars Express and Lion were included in each experiment as controls. In the four experiments a total of 3,146, 1,212, 2,591 and 4,895 plants were monitored, respectively. The plants were grown in individual pots with the genotypes arranged in a completely random design. In experiments 1–4 a total number of 100, 32, 82 and 98 DH lines were tested, respectively. In experiment 4 the two parents were also included. A total number of 214 different DH lines were tested in these four experiments. In experiment 4, randomly selected DH lines were retested: 52 from experiment 1, 15 from experiment 2 and 31 from experiment 3. The inoculations were carried out using a root-dip inoculation method described by Eynck et al. (2009a). The seedlings were grown at 22 °C in a greenhouse with a light regime of 14 h per day. Disease scoring was carried out weekly for disease symptoms over a time period of 4 weeks using a 9-scale assessment key (Eynck et al. 2009a). Area under the disease progress curve (AUDPC) values were calculated from the disease severity values. Growth reduction was calculated 28 days post-inoculation from the difference in the means of the height (in cm) from 20–24 non-inoculated and 20–24 inoculated plants.

Simple sequence repeat (SSR) marker analysis

Genomic DNA samples from the DH lines were isolated from fresh or freeze-dried leaf material according to Doyle and Doyle (1990). For all SSR analyses the M13-tailing procedure described by Berg and Olaisen (1994) was used. In this method the fluorescently labeled universal M13 primer 5′-AGGGTTTTCCCAGTCACGACGTT-3′ is added to the PCR reaction, and the forward primer of each SSR is appended with the sequence 5′-TTTCCCAGTCACGACGTT-3′. After the first cycle of amplification, the PCR fragments are subsequently amplified by the labeled universal primer. All SSR amplification products were separated using a LI-COR 4200 DNA Analyzer (LI-COR Biosciences) and scored visually. The SSR markers used in the present study derived from numerous published B. napus maps (Lowe et al. 2004; Suwabe et al. 2004; Piquemal et al. 2005; Qiu et al. 2006; Basunanda et al. 2007, 2010 Radoev 2007, 2008; Rygulla et al. 2008; Iniguez-Luy et al. 2009; Cheng et al. 2009; Nagaoka et al. 2010). See Supplementary Table S2 for detailed information on the SSR markers successfully mapped to the chromosomes of interest.

Map construction

Genetic mapping used a framework map published by Radoev et al. (2008) as a basis, with 191 markers mapped in 275 ExR53-DH lines. In addition, 15 polymorphic SSR markers mapping to chromosomes C1 and C5 were genotyped in 214 DH lines and added to the framework map using the program JoinMap 3.0 (Stam 1993). Of these 15 polymorphic SSR markers, 10 are shown in the genetic maps in Figs. 2 and 3 which are at least 1 cM distant from previously mapped SSR markers. Map distances measured in cM between markers were derived from Haldane’s function.

Statistical analyses and QTL mapping

QTL positions were localized using the software QGene, version 4.3.9 (Joehanes and Nelson 2008), by composite interval mapping with automatic forward cofactor selection allowing QTL cofactors instead of markers. The map was scanned at 2 cM intervals. A value of p ≤ 0.005 for Type I errors and a log10-likelihood ratio (LOD) value of 2.5 were used as criteria to indicate putative QTL positions. Estimates of the additive effect and phenotypic variation (R 2) explained by each QTL were computed. For each QTL, confidence intervals were calculated based on the 1-LOD drop-off method according to Lander and Botstein (1989). QTL were confirmed by re-analysis using the alternative software packages PlabQTL version 1.2 (Utz and Melchinger 1996), WindowsQTL Cartographer version 2.5 (Wang et al. 2011) and R/qtl (Broman et al. 2003). Calculation of broad-sense heritability was also performed using PlabQTL, while all other statistical analyses were carried out with SAS version 9.2 (SAS Institute Inc.).

RP-HPLC analysis

Phenolic compounds from hypocotyls were extracted with 80 % methanol using a method which was shown to provide the highest phenolic acid extraction efficiency from plant samples (Padda and Picha 2007). Extraction was performed from pooled samples of 24 plants according to Eynck et al. (2009b) from hypocotyls sampled 28 days after inoculation (38 days after sowing). Reverse phase–high performance liquid chromatography was performed using a Merck/Hitachi machine with a Nucleodur Sphinx RP column (Macherey–Nagel) and UV detection at 280 nm. The mobile phase consisted of water with 4.5 % formic acid (solvent A) and acetonitrile (solvent B). The elution was performed for 20 min with a 95/5, 10 min with a 75/25, followed by 20 min with a 2/98 mix of solvents A and B and a 1 ml/min flow rate, according to Lipsa et al. (2012). Peaks putatively corresponding to p-coumaric acid, caffeic acid, ferulic acid and sinapic acid were identified by comparison of the retention time with HPLC quality external standards obtained from Sigma-Aldrich and further confirmed by co-injection with standards. Further fractionation of the phenolic compounds from the 80 % methanolic extract of the hypocotyl from selected samples, via solid phase extraction C18 mini-columns (Alltech) into fractions representing phenolic acids, monomeric and oligomeric proanthocyanidines and polymeric proanthocyanidine/anthocyanidines (Lipsa et al. 2012), revealed that the majority of peaks were in the phenolic acid fraction and only one of approximately 45 peaks was also present in the monomeric and oligomeric proanthocyanidine fractions.

Results

Phenotypic evaluation of V. longisporum resistance

Mean values for the area under the disease progress curve (AUDPC, n = 20–24) ranged from 21.5 to 33.3 for the partially resistant reference cultivars Express and Lion, and from 27.8 to 69.6 for the highly susceptible reference cultivars Falcon and Laser in four different experiments. For these repeated measures for the four reference cultivars, multiple pairwise comparison showed significant differences between mean AUDPC values for some of the four experiments [Fig. 1A, p ≤ 0.05, least squares difference (LSD) test].

Fig. 1
figure 1

Mean area under the disease progress curve (AUDPC) and mean growth reduction values for four reference cultivars inoculated with V. longisporum in four greenhouse screening experiments. Bars show the standard error of the mean calculated from 20 to 24 single plants. Alphabetic annotations show significant differences based on pairwise comparison for each cultivar (A–H, LSD post hoc test, p ≤ 0.05)

Full size image

Based on the reaction of the reference cultivars, the disease pressure in experiment 2 was lower than in the other three experiments, resulting in up to 60 % lower mean AUDPC values (white bars in Fig. 1A). Similar results were also found for the mean growth reduction values between these four experiments (Fig. 1B). Differences between experiments for the mean AUDPC and growth reduction values were also significant when including all genotypes in the analysis (p ≤ 0.001, LSD test).

This could be due to differences in the accuracy of visual disease assessment, or to environmental effects (e.g. temperature and humidity fluctuations between greenhouse experiments). Because all visual resistance scoring was done by the same person using a standardized scheme, the observed variations between the four experiments are considered to be due to differences in disease pressure rather than inaccuracies in disease scoring. The interaction between varieties and experiments for AUDPC and growth reduction was significant (p ≤ 0.001). Because a standardized inoculum and a standardized scoring scheme were used in all four experiments, these differences between experiments are most likely due to environment differences. Accordingly, the four greenhouse screening experiments were treated as different environments and used for four independent QTL analyses instead of one combined QTL analysis.

Phenotypic variation in the AUDPC and the degree of disease-induced growth reduction in the DH mapping population Express 617 × R53 (ExR53-DH) revealed similar, continuous frequency distributions in the four resistance screening experiments with different subsets of the population. In experiment 4, randomly selected DH lines from experiments 1 to 3 were retested. In all four experiments the distributions for both normalized AUDPC (Supplementary Figure S3) and growth reduction (data not shown) were slightly skewed towards high susceptibility.

In greenhouse experiment 4, 52 DH lines (53 %) showed a lower mean AUDPC than the resistant parent R53, while eight DH lines (8 %) had a higher mean AUDPC than the parent Express 617 (Supplementary Figure S3). Estimates of broad-sense heritability (h 2) were high, ranging from 0.73 to 0.87 for both resistance-related traits in the different subsets of DH lines retested in experiment 4 and calculated from the four experiments. This high heritability was confirmed by significant Pearson correlation coefficients (R = 0.59–0.82) between the results of the independent greenhouse phenotyping experiments for different populations or subpopulations.

High correlations (R = 0.87, 0.67, 0.89 and 0.81; significant at p ≤ 0.01) were observed between leaf symptoms and the degree of growth reduction in infected plants from experiments 1 to 4, respectively.

QTL for V. longisporum resistance

In the three independent resistance tests 1, 3 and 4, significant and stable major QTL with LOD scores larger than 2.5 for both resistance-related traits, AUDPC and growth reduction, were found at the same position on chromosome C5 (Fig. 2). Resistance tests 1 and 3 each comprised different subsets of the ExR53-DH population; in resistance test 4 some randomly selected ones were retested. When analyzed using composite interval mapping (CIM) with forward cofactor selection in the software QGENE, this QTL explained between 16 and 34 % of the phenotypic variance for the different traits and datasets (Supplementary Table S3). Similar results were obtained using different CIM approaches implemented in the software packages PlabQTL, R/qtl and Windows QTL Cartographer (data not shown).

Fig. 2
figure 2

Comparison of quantitative trait loci for V. longisporum resistance-related traits and soluble phenolic metabolites in the hypocotyl localized on chromosome C5. Blocks indicate confidence intervals of the QTL. Data for the population SW99-307 are from Rygulla et al. (2008). Marker alleles BRMS030_210 and Ra2F11_230 are designated BRMS030 and Ra2F11_b, respectively, in Rygulla et al. (2008). Exp experiment

Full size image

Besides the major QTL on chromosome C5, an additional minor QTL was identified on chromosome C1 for AUDPC values in experiments 1 and 3, and for growth reduction in experiment 3 (Fig. 3, Supplementary Table S3). No significant resistance-related QTL were detected in experiment 2, presumably due to the low number of DH lines tested (n = 32). No clear evidence for epistatic interactions between digenic loci or main effect QTL was found using PlabQTL or R/qtl.

Fig. 3
figure 3

Comparison of quantitative trait loci for V. longisporum resistance and soluble phenolic metabolites in the hypocotyl localized on chromosome C1. Blocks indicate confidence intervals of the QTL. Exp experiment

Full size image

Rygulla et al. (2008) detected two major QTL for V. longisporum resistance on C4 and C5 in the mapping population SW99-307. Mapping of the SSR markers Ra2F11_b and BRMS030, which flank one of the major QTL on C5 in the SW99-307 DH population, revealed that they also flank the major QTL on C5 in ExR53-DH (Fig. 2; here called Ra2F11_230 and BRMS030_210 according to the allele sizes in bp).

Combined effect of resistance alleles from two QTL and both parents

The allele contributing resistance at the detected major QTL on C5 was derived from the C genome donor of the resistant resynthesized rapeseed parent R53, whereas the allele contributing resistance at the minor QTL on C1 originates from the inbred line Express 617 (Supplementary Table S3). Express 617 was derived by repeated self-pollination from the partially resistant cv. Express, which is considered to exhibit a medium field resistance towards V. longisporum infection. The mean normalized AUDPC values of V. longisporum-infected DH lines drop about 40 % from an average of 0.98 for lines containing no favorable allele to 0.58 for lines containing two favorable alleles at the two QTL regions on C1 and C5 from both parents (Table 1). Similarly, the mean normalized growth reduction upon V. longisporum infection compared to mock-inoculated DH lines is about 33 % in DH lines with both favorable alleles in these QTL regions, compared to about 50 % without the two favorable alleles from the respective parents.

Table 1 Disease symptoms (AUDPC, growth reduction) in ExR53-DH lines with different combinations of resistance alleles from the two V. longisporum resistance QTL on chromosomes C1 and C5 from Express 617 and R53, respectively
Full size table

QTL validation

From 55 publicly available SSR markers mapped on chromosomes C1 and C5 in 18 other mapping populations, 13 markers are located 10 cM either side of the QTL peaks in ExR53-DH. The percentages of phenotypic variation explained by three selected marker alleles derived from the QTL regions on C1 and C5 with mean AUDPC values in the two mapping populations and in four genetically diverse DH populations are shown in Table 2.

Table 2 The percentage of variation explained based on simple linear regression of marker alleles from V. longisporum resistance QTL regions and disease scores in genetically diverse DH populations
Full size table

Two of the DH populations used for QTL validation tested together in one resistance screening are genetically similar half-sister populations, SW08-190001 and SW08-190002, which only differ in the C genome resistance donor (BRA 1008 or HRI 8207) of their respective resynthesized rapeseed parent (Supplementary Table S1). Although on average these DH populations should be about 75 % genetically identical, their normalized AUDPC values in fact show contrasting associations with different marker alleles (Table 2). Whereas SW08-190001 significantly associates only with marker Ol10E12 derived from the minor QTL on C1 in one of the two experiments, SW08-190002 shows a significant association only with the marker CB10065 derived from the major QTL region on C5 in one of the two experiments. This result suggests that the markers allow differentiation between different C genome resistance donors and the resistance QTL on C1 and C5. Comparison of the marker–AUDPC association of SW08-190002 with population DSV-1575, which is genetically diverse but shares the same C genome resistance donor HRI 8207, confirms that the marker CB10065 (derived from the C5 major QTL) is also significantly related to resistance. This concurs with the pedigree information in Supplementary Table S1 and confirms that CB10065 can be effectively used to select for this resistance QTL in different genetic backgrounds.

Co-localization of QTL for V. longisporum resistance and metabolites from the phenylpropanoid pathway

The total soluble phenolics concentration in the hypocotyl of 98 DH lines from the mapping population measured by RP-HPLC increased for every single DH line 1.2-fold up to 5-fold in experiment 4 upon infection with V. longisporum. Supplementary Figure S4 shows the frequency distributions for the total soluble phenolics concentration in the hypocotyls of 98 DH lines for the V. longisporum-inoculated compared to the mock-inoculated set in experiment 4.

Significant but weak correlation (R 2 = 4 %) was observed between total phenolics concentration and AUDPC in the V. longisporum-inoculated DH line set (n = 98), but not in the mock-inoculated DH line set. For the individual phenolic acids, 34 QTL with a LOD score threshold of 2.5 were detected for a total number of 43 RP-HPLC peaks from mock-inoculated DH lines, and 37 QTL for 49 peaks from V. longisporum-inoculated lines. Four major regions were identified in the DH line set for the mock-inoculation treatment on chromosomes A9, C1, C6 and C8 showing overlapping confidence intervals for at least three QTL involved in the concentrations of individual phenolic acids in the hypocotyl. In the DH line set for the V. longisporum-inoculation treatment, two major regions were identified on chromosomes C1 and C5 (for details see Supplementary Table S4).

Further analysis focused on the phenylpropanoid compounds that exhibited a significant correlation with AUDPC and explained at least 5 % of the phenotypic variation in AUDPC. These were 13 compounds from the V. longisporum-inoculated DH line data set and five compounds from the mock-inoculated DH line data set (see Supplementary Table S4). Of these 18 phenylpropanoid compounds significantly correlated with AUDPC, seven produced QTL with a LOD score of 2.5 which showed at least one QTL co-localizing with the QTL for AUDPC on chromosome C1 or C5 (summarized in Supplementary Table S3, detailed data in Supplementary Table S4, co-localizing QTL labelled in green). Six of these soluble phenylpropanoid compounds where QTL also co-localized with the QTL for AUDPC on C1 or C5 (Figs. 2, 3) were detected only in the V. longisporum-inoculated data set (Table 3). These six compounds explained 7–37 % of the phenotypic variation in AUDPC. In contrast, only one phenylpropanoid compound identified as caffeic acid, where a QTL for this compound also co-localized with the QTL for AUDPC on C1 (Table 3, Supplementary Table S3), was detected in the mock-inoculated data set. In this case, 15 % of the phenotypic variation of AUDPC was explained by the variation in caffeic acid concentration in the mock-inoculated data set (Table 3).

Table 3 QTL for soluble phenolic compounds in the hypocotyls of the mock- and V. longisporum-inoculated mapping population ExR53-DH
Full size table

Correlations of constitutive or induced concentrations of the upstream lignin biosynthesis precursors p-coumaric acid, ferulic acid and sinapic acid with AUDPC were not observed. In contrast, the constitutive concentration of the lignin precursor caffeic acid significantly correlated with AUDPC, and a QTL for the constitutive expression of caffeic acid in the hypocotyl co-localized with the minor QTL for V. longisporum resistance on C1 (retention time 11.3 min; Fig. 3; Table 3). Another two, as yet unidentified, induced soluble phenolic compounds correlated with AUDPC and co-localized with the V. longisporum QTL on C1 (retention times 9.7 and 24.5 min; Fig. 3; Table 3). Furthermore, five other soluble phenolic compounds that were induced upon V. longisporum infection also correlated with resistance and co-localized with the major resistance QTL on C5 (Fig. 2; Table 3).

These results provide preliminary evidence for regulatory changes in the phenylpropanoid biosynthesis pathway as a potential component of resistance to V. longisporum infection.

Discussion

Verticillium disease is an increasing threat to oilseed rape production in Europe. Resistance to V. longisporum infection in B. napus has been described as being polygenic or multifactorial (Happstadius et al. 2003; Rygulla et al. 2007a, b, 2008; Eynck et al. 2009a). In an earlier study we identified major QTL for resistance on chromosomes C4 and C5 and minor QTL on A6 and C8 in a DH mapping population, SW99-307, produced from resynthesized rapeseed (Rygulla et al. 2008). In the current study a different DH population, ExR53-DH, was used for resistance mapping to dissect the genetic basis of V. longisporum resistance present in commercial and resynthesized rapeseed material. This was approached by genetic analysis using a cross of an inbred line from a commercial cultivar exhibiting partial resistance, Express 617, and a resynthesized rapeseed line, R53, of entirely different origin exhibiting pronounced resistance.

High estimates of broad-sense heritability (h 2) for both major resistance-related traits, AUDPC and growth reduction, showed that the applied greenhouse test procedure is useful for confidently selecting resistant progenies in breeding, and for accurate genetic mapping. Screening for resistance against V. longisporum is extremely time-consuming and costly, however, making it impracticable to implement in a breeding program. On the other hand, phenotypic correlations between the AUDPC values normalized based on the reaction of reference cultivars from different greenhouse experiments, and between greenhouse and field experiments (data not shown), suggest that QTL influencing resistance to artificial inoculation in the greenhouse are highly relevant for field resistance against V. longisporum. Effective molecular markers derived from resistance QTL are therefore an extremely valuable resource for breeding of new oilseed rape cultivars with effective field resistance against this important disease.

High correlations between leaf symptoms and degree of growth reduction in infected plants confirm previous observations from greenhouse resistance tests, where a high correlation has been documented between AUDPC and stunted growth for V. longisporum-infected B. napus accessions (Eynck et al. 2009a). On the other hand, Verticillium-induced stunting in A. thaliana under greenhouse conditions appears to be genotype-dependent (Floerl et al. 2010; Veronese et al. 2003; Häffner et al. 2010; Steventon et al. 2001), while B. rapa accessions seem to show a weaker correlation between AUDPC and stunting than B. napus (Eynck et al. 2009a). Furthermore, V. longisporum-induced stunting has never been observed in B. napus infested in the fields excluding relative growth reduction as a suitable parameter for scoring of field resistance.

The continuous frequency distributions in four resistance screening experiments with different subsets of the ExR53-DH population confirm the quantitative inheritance of V. longisporum resistance. On the other hand, this population exhibited a transgressive segregation for resistance that was demonstrated to be due to recombination of additive alleles contributed by both parents, the inbred line Express 617 and the resynthesized rapeseed line R53, without clear evidence for epistatic interaction. This underlines the value of effective genetic markers for combination of different resistance QTL from different genetic backgrounds, including less resistant genotypes.

In an earlier QTL analysis using the mapping population SW99-307, the resistance was derived from a white cabbage (B. oleracea ssp. oleracea convar. capitata), which had been crossed with a low erucic acid winter turnip rape (B. rapa ssp. oleifera) to generate a synthetic B. napus in Sweden in 1976 (Rygulla et al. 2008). In contrast, the resistance donor R53 was produced at Göttingen University from a cross between a kale (B. oleracea ssp. oleracea convar. acephela) and a chinese cabbage (B. rapa ssp. pekinensis) (Radoev et al. 2008). Markers derived from R53 that flank the major QTL for V. longisporum resistance on C5 also flank the major QTL for the same trait found on C5 in the former study. This chromosome region therefore appears to play an important role in the expression of V. longisporum resistance from very different C genome genetic backgrounds, so that molecular markers linked to resistance alleles in this region might be broadly applicable in marker-assisted breeding for V. longisporum resistance. Combination with markers linked to the minor resistance QTL on chromosome C1, which represents yet another genetic origin for resistance against V. longisporum, underlines the potential for a marker-assisted pyramiding strategy to assist the effective combination of V. longisporum resistance from resynthesized rapeseed donors with existing resistance in elite breeding lines.

On the other hand, QTL and associated markers identified in biparental mapping populations are rarely directly useful in marker-assisted selection, requiring verification that they are effective in different genetic backgrounds. Due to limited polymorphism of individual QTL-linked markers in different genotypes, a suite of additional markers within a narrow window spanning the QTL are often required for QTL validation and establishment of a broadly applicable marker-assisted selection strategy (Akhtar et al. 2010). In the present study a QTL validation experiment showed that a number of selected QTL-derived markers were polymorphic in numerous, genetically diverse breeding populations. These markers now allow breeders to distinguish the common resistance QTL on chromosomes C1 and C5 and combine them into new oilseed rape breeding lines and cultivars with increased quantitative resistance to V. longisporum.

The resistance mechanism active against V. longisporum in rapeseed is largely unknown. However, histochemical analysis of a susceptible and a resistant B. napus genotype revealed that the phenol metabolism plays a crucial role in defence against V. longisporum and that these defence responses are located in the vascular tissues of plant hypocotyls, leading to the accumulation of soluble phenolics and deposition of cell wall-bound phenolics and lignin (Eynck et al. 2009b). Therefore, resistance to V. longisporum was not only studied by measuring AUDPC, but also by quantifying the concentration of soluble phenolic compounds in the hypocotyl of the mapping population in the mock- and V. longisporum-inoculated treatment.

By correlation analysis, genetic mapping and QTL analysis, the empiric observations for the total phenolic concentrations in the hypocotyls of resistant and susceptible lines published by Eynck et al. (2009b) were confirmed. In addition, it was shown in more detail that some individual phenylpropanoid compounds expressed in the hypocotyl of the mapping population and measured by HPLC exhibited a weak to medium correlation with V. longisporum resistance in B. napus. Some of the detected QTL regions for these phenylpropanoid concentrations also co-localized in genomic regions together with the major and minor resistance QTL.

Over 30 QTL were detected for soluble phenolic compounds in the hypocotyl of the mapping population in the mock- as well as in the V. longisporum-inoculated data set. In only about half of the cases, the QTL were located in the same genomic region, suggesting a strong V. longisporum-induced activation of different genomic regions for synthesis or modification of soluble phenolic compounds. There were four major regions (on A9, C1, C6 and C8) involved in constitutive soluble phenolic compound concentrations and two major regions (on C1 and C5) involved in induced soluble phenolic compound concentrations. One major region on C1 was detected in the mock- as well as in the V. longisporum-inoculated data set. Data on the genetic mapping of phenolic acids in rapeseed is limited. However, a QTL for acid detergent lignin (ADL) concentration in seeds within the same major region on C5 was mapped in the sister DH population Express 617 × 1012-98 (data not shown). Also, two QTL for phenolic compounds, sinapic acid and sinapoyl-glucose, have been mapped in seeds within the major region on C1 (data not shown) and the gene UDP-glucose:sinapate glucosyltransferase has been mapped in the same region (Mittasch et al. 2010), also supporting that these regions on C1 and C5 are major genomic regions involved in phenylpropanoid synthesis or modification in rapeseed.

Of a total of 71 QTL for phenylpropanoid compounds, 18 co-localized with the major and minor QTL for V. longisporum resistance on C1 or C5. Less than half of these phenylpropanoid compounds (seven out of 18) also showed significant correlations of their concentrations with AUDPC, suggesting that they do not co-localize by chance, but might be associated with resistance. Most of these co-localizing QTL where in addition the concentrations of the respective phenylpropanoid compounds are significantly correlated with resistance resulting in an R 2 of more than 5 % were detected in the V. longisporum-inoculated data set (six out of seven). The variation in concentration of these six phenylpropanoid compounds explained 7–37 % of the phenotypic variation in AUDPC. This indicates that, for most of these phenylpropanoid compounds, concentrations are changed in the DH lines upon V. longisporum infection. Of these six phenylpropanoid metabolites, the concentrations of two were negatively correlated with AUDPC, meaning higher concentrations in resistant lines upon infection. Concentrations of the four other compounds were positively correlated with AUDPC, meaning higher concentrations in susceptible lines upon infection, including one compound which explained 37 % of the phenotypic variation in AUDPC. These compounds might not be directly involved in a resistance reaction, but rather might be precursors of compounds involved in resistance. However, the phenolic compound which reveals the second largest R 2 value, explaining 15 % of the phenotypic variation in AUDPC values, is constitutively expressed in the hypocotyl. This compound was identified by co-migration with an external standard as caffeic acid (RT 11.3). The QTL for the constitutive caffeic acid concentration in the hypocotyls co-localized with the minor resistance QTL on C1. Caffeic acid is a key component of the phenylpropanoid pathway and major lignin precursor (e.g. Dixon and Reddy 2003). Resistant lines showed a lower constitutive concentration of caffeic acid than susceptible lines. This might suggest that caffeic acid was more effectively incorporated in resistant lines into other phenolic compounds or cellular structures which are part of preformed physical or physiological barriers to infection, e.g. in lignin of cell walls within the vascular system. This is in agreement with biochemical and microscopical investigations of a susceptible and resistant B. napus line by Eynck et al. (2009b), who described that at earlier time points of infection preformed soluble and cell wall-bound phenolics appear to limit the extent of infection and colonization by the fungal pathogen, whereas de novo formation of lignin and lignin-like polymers becomes more important at later stages of infection.

Resistance of plants to plant pathogens is often a multifactorial process. The accumulation of low-molecular-weight phenols leading to the formation of biopolymers that restrict the spread of the pathogens (e.g. lignin and callose) is only one part of the diverse layers of plant responses to pathogen infection. Low-molecular-weight phenols accumulate in many host–pathogen systems in both resistant and susceptible interactions early after infection. These phenolic compounds, including caffeoyl glucose and caffeoyl esters, might ultimately serve as precursors for compounds essential to expression of resistance. These compounds might also be associated with cellular browning and tissue necrosis that may result from oxidation and polymerization of O-diphenols (Nicholson and Hammerschmidt 1992). However, enhanced lignification and differences in lignin composition have been described as being the main resistance pathway to fungal invasion in a number of different host–pathogen interactions (reviewed by Nicholson and Hammerschmidt 1992; Dixon and Paiva 1995; Vance et al. 1980). Lignification is not only an induced mechanism in the resistance of plants to fungal pathogens, but also plays an important role as a constitutive resistance component (e.g. Eynck et al. 2012, resistance of C. sativa to S. sclerotiorum). Phenolic acids and phenylpropanoid metabolites have also been assumed to be implicated in the interaction of V. longisporum and A. thaliana. In a study by Götze et al. (2011) it was shown that leaves of A. thaliana accumulate phenylpropanoids and lignans soon after V. longisporum root infection. An activation of phenylpropanoid metabolism-related genes and lignin deposition in the hypocotyl after inoculation of cotton with V. dahliae was reported from a study based on the global gene regulation analysis of phenylpropanoid metabolism-related genes using RNA-Seq, in combination with histochemical analyses of a resistant and susceptible cotton genotype (Xu et al. 2011).

The V. longisporum-induced regulation of individual soluble phenolic acids in the hypocotyl of resistant genotypes, and their correlation and co-localization with QTL for resistance-related traits, described in this study indicates that genes from the phenylpropanoid pathway should be considered as promising candidate genes for expression of V. longisporum resistance in oilseed rape. Identification of the as yet unidentified HPLC peaks correlating with V. longisporum resistance and co-localizing with resistance QTL by mass spectroscopy will allow the identification of further putative genes and the evaluation of which role they play in this host–pathogen interaction.